Search

Abstract: Geostatistical simulation techniques have become a widely used tool for the modeling of oil and gas reservoirs and the assessment of uncertainty. The Direct Sampling (DS) algorithm is a recent multiple-point statistical simulation technique. It directly samples the training image (TI) during the simulation process by calculating distances between the TI patterns and the given data events found in the simulation grid (SG). Omitting the prior storage of all the TI patterns in a database, the DS algorithm can be used to simulate categorical, continuous and multivariate variables. Three fundamental input parameters are required for the definition of DS applications: the number of neighbors n, the acceptance threshold t and the fraction of the TI to scan f. For very large grids and complex spatial models with more severe parameter restrictions, the computational costs in terms of simulation time often become the bottleneck of practical applications. This paper focuses on an innovative implementation of the Direct Sampling method which exploits the benefits of graphics processing units (GPUs) to improve computational performance. Parallel schemes are applied to deal with two of the DS input parameters, n and f. Performance tests are carried out with large 3D grid size and the results are compared with those obtained based on the simulations with central processing units (CPU). The comparison indicates that the use of GPUs reduces the computation time by a factor of 10X–100X depending on the input parameters. Moreover, the concept of the search ellipsoid can be conveniently combined with the flexible data template of the DS method, and our experimental results of sand channels reconstruction show that it can improve the reproduction of the long-range connectivity patterns. [Copyright &y& Elsevier]

In this paper we develop a graphics processing unit (GPU)-based massively parallel approach for efficient computation of electromagnetic scattering via a proposed double-layer vegetation model composed of vegetation and ground layers. The proposed vector radiative transfer (VRT) model for vegetation scattering considers different sizes and orientations of the leaves. It uses the Monte Carlo method to calculate the backward scattering coefficients of rough ground and vegetation where the leaves are approximated as a large number of randomly oriented flat ellipsoids and the ground is treated as a Gaussian random rough surface. In the original CPU-based sequential code, the Monte Carlo simulation to calculate the electromagnetic scattering of vegetation takes up 97.2% of the total execution time. In this paper we take advantage of the massively parallel compute capability of NVIDIA Fermi GTX480 with the Compute Unified Device Architecture (CUDA) to compute the multiple scattering of all the leaf groups simultaneously. Our parallel design includes the registers for faster memory access, the shared memory for parallel reduction, the pipelined multiple-stream asynchronous transfer, the parallel random number generator and the CPU-GPU heterogeneous computation. By using these techniques, we achieved speedup of 213-fold on the NVIDIA GTX 480 GPU and 291-fold on the NVIDIA GTX 590 GPU as compared with its single-core CPU counterpart. [ABSTRACT FROM PUBLISHER]

The processing speed for positron emission tomography (PET) image reconstruction has been greatly improved in recent years by simply dividing the workload to multiple processors of a graphics processing unit (GPU). However, if this strategy is generalized to a multi-GPU cluster, the processing speed does not improve linearly with the number of GPUs. This is because large data transfer is required between the GPUs after each iteration, effectively reducing the parallelism. This paper proposes a novel approach to reformulate the maximum likelihood expectation maximization (MLEM) algorithm so that it can scale up to many GPU nodes with less frequent inter-node communication. While being mathematically different, the new algorithm maximizes the same convex likelihood function as MLEM, thus converges to the same solution. Experiments on a multi-GPU cluster demonstrate the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

A detailed scheme of mapping Precorrected-FFT (P-FFT) method to graphics processing units (GPUs) is presented through the solution of electric field integral equation (EFIE) for modeling the scattering from 3-D perfectly electric conducting (PEC) object. The procedure of the P-FFT method is analyzed to identify its computationally intensive spots, which are to be implemented on GPU. Based on this, a mapping scheme is proposed. Moreover, the optimization of the GPU code is discussed to demonstrate its importance in achieving better performance. The optimization scheme is broken down into several stages to show the gradual performance improvement in each stage. Accurate numerical results obtained using the GPU code validate the proposed mapping and optimization scheme. The performance of GPU code is benchmarked and the acceleration ratio achieved is more than 50. [ABSTRACT FROM AUTHOR]

Abstract: This paper describes sum of absolute differences (SAD) algorithm to calculate optical flow function. This algorithm is used for optical flow estimation between two frames of colored photo image. The algorithm is combined with parallel processing, modern general purpose Compute Unified Device Architecture (CUDA) Graphic Processing Unit (GPU) to reduce significantly the time consumption of computation of using traditional Central Processing Unit (CPU). This algorithm is tested on real world colored photo image sequences. The performance in term of time consumption of algorithm using parallel processing (using GPU) is then compared to serial processing (using CPU). [Copyright &y& Elsevier]

Abstract: Several areas of knowledge are being benefited with the reduction of the computing time by using the technology of graphics processing units (GPU) and the compute unified device architecture (CUDA) platform. In case of evolutionary algorithms, which are inherently parallel, this technology may be advantageous for running experiments demanding high computing time. In this paper, we provide an implementation of a co-evolutionary differential evolution (DE) algorithm in C-CUDA for solving min–max problems. The algorithm was tested on a suite of well-known benchmark optimization problems and the computing time has been compared with the same algorithm implemented in C. Results demonstrate that the computing time can significantly be reduced and scalability is improved using C-CUDA. As far as we know, this is the first implementation of a co-evolutionary DE algorithm in C-CUDA. [Copyright &y& Elsevier]

Abstract: Geostatistical simulations have become a widely used tool for modeling of oil and gas reservoirs and the assessment of uncertainty. One important current issue is the development of high-resolution models in a reasonable computational time. A possible solution is based on taking advantage of parallel computational strategies. In this paper we present a new methodology that exploits the benefits of graphics processing units (GPUs) along with the master–slave architecture for geostatistical simulations that are based on random paths. The methodology is a hybrid method in which different levels of master and slave processors are used to distribute the computational grid points and to maximize the use of multiple processors utilized in GPU. It avoids conflicts between concurrently simulated grid points, an important issue in high-resolution and efficient simulations. For the sake of comparison, two distinct parallelization methods are implemented, one of which is specific to pattern-based simulations. To illustrate the efficiency of the method, the algorithm for the simulation of pattern is adapted with the GPU. Performance tests are carried out with three large grid sizes. The results are compared with those obtained based on simulations with central processing units (CPU). The comparison indicates that the use of GPUs reduces the computation time by a factor of 26–85. [Copyright &y& Elsevier]

The Weather Research and Forecasting (WRF) model is a next-generation mesoscale numerical weather prediction system. It is designed to serve the needs of both operational forecasting and atmospheric research for a broad spectrum of applications across scales ranging from meters to thousands of kilometers. Microphysics plays an important role in weather and climate prediction. Microphysics includes explicitly resolved water vapor, cloud, and precipitation processes. Several bulk water microphysics schemes are available within the WRF, with different numbers of simulated hydrometeor classes and methods for estimating their size, fall speeds, distributions and densities. Stony Brook University scheme is a 5-class scheme with riming intensity predicted to account for the mixed-phase processes. In this paper, we develop an efficient Graphics Processing Unit (GPU) based Stony Brook University scheme. The GPU-based Stony Brook University scheme was compared to a CPU-based single-threaded counterpart on a computational domain of 422 \times 297 horizontal grid points with 34 vertical levels. The original Fortran code was first rewritten into a standard C code. After that, C code was verified against Fortran code and CUDA C extensions were added for data parallel execution on GPUs. On a single GPU, we achieved a speed-up of 213\times with data I/O and 896\times without I/O on NVIDIA GTX 590. Using multiple GPUs, a speed-up of 352\times is achieved with I/O for 4 GPUs. We will also discuss how data I/O will be less cumbersome if we ran the complete WRF model on GPUs. [ABSTRACT FROM PUBLISHER]

Next-generation mesoscale numerical weather prediction system, the Weather Research and Forecasting (WRF) model, is a designed for dual use for forecasting and research. WRF offers multiple physics options that can be combined in any way. One of the physics options is radiance computation. The major source for energy for the earth's climate is solar radiation. Thus, it is imperative to accurately model horizontal and vertical distribution of the heating. Goddard solar radiative transfer model includes the absorption duo to water vapor, \O3, \O2, \CO2, clouds and aerosols. The model computes the interactions among the absorption and scattering by clouds, aerosols, molecules and surface. Finally, fluxes are integrated over the entire shortwave spectrum from 0.175 \mu\m to 10 \mu\m. In this paper, we develop an efficient graphics processing unit (GPU) based Goddard shortwave radiative scheme. The GPU-based Goddard shortwave scheme was compared to a CPU-based single-threaded counterpart on a computational domain of 422\times297 horizontal grid points with 34 vertical levels. Both the original FORTRAN code on CPU and CUDA C code on GPU use double precision floating point values for computation. Processing time for Goddard shortwave radiance on CPU is 22106 ms. GPU accelerated Goddard shortwave radiance on 4 GPUs can be computed in 208.8 ms and 157.1 ms with and without I/O, respectively. Thus, the speedups are 116\times with data I/O and 141\times without I/O on two NVIDIA GTX 590 s . Using single precision arithmetic and less accurate arithmetic modes the speedups are increased to 536\times and 259\times, with and without I/O, respectively. [ABSTRACT FROM PUBLISHER]

This paper presents a novel parallel architecture for accelerating quadrature methods used for pricing complex multi-dimensional options, such as discrete barrier, Bermudan and American options. We explore different designs of the quadrature evaluation core including optimized pipelined hardware designs in reconfigurable logic and a compute unified device architecture (CUDA)-based graphics processing unit (GPU) design. A parametrizable automated system is presented for generating hardware quadrature evaluation cores with an arbitrary number of dimensions. The performance and energy consumption of field-programmable gate arrays (FPGAs), GPUs, and central processing units (CPUs) are compared across different number of dimensions and precisions. Our evaluation shows that the 100 MHz Virtex-4 xc4vlx160 FPGA design is 4.6 times faster and 25.9 times more energy efficient than a multi-threaded optimized software implementation running on a Xeon W3504 dual-core CPU. It is also 2.6 times faster and 25.4 times more energy efficient than a GPU with comparable silicon process technology. [ABSTRACT FROM AUTHOR]

A novel approach for high-speed measurement of surface grid line intersection points across multiple video frames is described. Grids are printed or etched onto otherwise featureless surfaces for applications that include 3-D surface reconstruction, sheet metal surface strain measurement, and others. To achieve the necessary subpixel location accuracy, close-range imaging is used with data collected by a hand-guided digital video camera mounted on a portable articulated arm coordinate measuring machine. Grid extraction is based on ridge detection in a parallelized scale space, implemented with a 480-core graphical processing unit (GPU). The close-range narrow-depth-of-field focus variations within the video sequence are intrinsically handled by the scale space. Ridge linking, filtering, and parabola fitting are used to accurately extract the grid intersection points. While computationally intensive, experimental implementation using the parallel GPU hardware has achieved sustained throughput exceeding 15 frames per second, with more than 100 intersections extracted per frame. Experimental results are presented for both synthetically generated and actual video sequences. [ABSTRACT FROM PUBLISHER]

The variable preconditioned (VP) Krylov subspace method with mixed precision is implemented on graphics processing unit (GPU) using compute unified device architecture (CUDA), and the linear system obtained from the edge element is solved by means of the method. The VPGCR method has the sufficient condition for the convergence. This sufficient condition leads us that the residual equation for the preconditioned procedure of VPGCR can be solved in the range of single precision. To stretch the sufficient condition, we propose the hybrid scheme of VP Krylov subspace method that uses single and double precision operations. The results of computations show that VPCG with mixed precision on GPU demonstrated significant achievement than that of CPU. Especially, VPCG-JOR on GPU with mixed precision is 41.853 times faster than that of VPCG-CG on CPU. [ABSTRACT FROM PUBLISHER]

In lung cancer screening, benign and malignant nodules can be classified through nodule growth assessment by the registration and, then, subtraction between follow-up computed tomography scans. During the registration, the volume of nodule regions in the floating image should be preserved, whereas the volume of other regions in the floating image should be aligned to that in the reference image. However, ground glass opacity (GGO) nodules are very elusive to automatically segment due to their inhomogeneous interior. In other words, it is difficult to automatically define the volume-preserving regions of GGO nodules. In this paper, we propose an accurate and fast nonrigid registration method. It applies the volume-preserving constraint to candidate regions of GGO nodules, which are automatically detected by gray-level cooccurrence matrix (GLCM) texture analysis. Considering that GGO nodules can be characterized by their inner inhomogeneity and high intensity, we identify the candidate regions of GGO nodules based on the homogeneity values calculated by the GLCM and the intensity values. Furthermore, we accelerate our nonrigid registration by using Compute Unified Device Architecture (CUDA). In the nonrigid registration process, the computationally expensive procedures of the floating-image transformation and the cost-function calculation are accelerated by using CUDA. The experimental results demonstrated that our method almost perfectly preserves the volume of GGO nodules in the floating image as well as effectively aligns the lung between the reference and floating images. Regarding the computational performance, our CUDA-based method delivers about 20× faster registration than the conventional method. Our method can be successfully applied to a GGO nodule follow-up study and can be extended to the volume-preserving registration and subtraction of specific diseases in other organs (e.g., liver cancer). [ABSTRACT FROM AUTHOR]

General purpose computing on graphical processing units (GPGPU) is a paradigm shift in computing that promises a dramatic increase in performance. GPGPU also brings an unprecedented level of complexity in algorithmic design and software development. In this paper, we present an efficient parallel fault simulator, FSimGP^2, that exploits the high degree of parallelism supported by a state-of-the-art graphic processing unit (GPU) with the NVIDIA compute unified device architecture. A novel 3-D parallel fault simulation technique is proposed to achieve extremely high computation efficiency on the GPU. Global communication is minimized by concentrating as much work as possible on the local device's memory. We present results on a GPU platform from NVIDIA (a GeForce GTX 285 graphics card) that demonstrate a speedup of up to 63\times and 4\times compared to two other GPU-based fault simulators and up to 95\times over a state-of-the-art algorithm on conventional processor architectures. [ABSTRACT FROM AUTHOR]

In this paper, we develop a novel Graphics Processing Unit (GPU)-based high-performance Radiative Transfer Model (RTM) for the Infrared Atmospheric Sounding Interferometer (IASI) launched in 2006 onboard the first European meteorological polar-orbiting satellites, METOP-A. The proposed GPU RTM processes more than one profile at a time in order to gain a significant speedup compared to the case of processing just one profile at a time. The radiative transfer model performance in operational numerical weather prediction systems nowadays still limits the number of channels they can use in hyperspectral sounders to only a few hundreds. To take the full advantage of such high resolution infrared observations, a computationally efficient radiative transfer model is needed. Our GPU-based IASI radiative transfer model is developed to run on a low-cost personal supercomputer with 4 NVIDIA Tesla C1060 GPUs with total 960 cores, delivering near 4 TFlops theoretical peak performance. The model exhibited linear scaling with the number of graphics processing units. Computing 10 IASI radiance spectra simultaneously on a GPU, we reached 763x speedup for 1 GPU and 3024x speedup for all 4 GPUs, both with respect to the original single-threaded Fortran CPU code. The significant 3024x speedup means that the proposed GPU-based high-performance forward model is able to compute one day's amount of 1,296,000 IASI spectra within 6 minutes, whereas the original CPU-based version will impractically take more than 10 days. The GPU-based high-performance IASI radiative transfer model is suitable for the assimilation of the IASI radiance observations into the operational numerical weather forecast model. [ABSTRACT FROM PUBLISHER]

Modern active and passive satellite and airborne sensors with higher temporal, spectral and spatial resolutions for Earth remote sensing result in a significant increase in data volume. This poses a challenge for data transmission over error-prone wireless links to a ground receiving station. Low-density parity-check (LDPC) codes have been adopted in modern communication systems for robust error correction. Demands for LDPC decoders at a ground receiving station for efficient and flexible data communication links have inspired the usage of a cost-effective high-performance computing device. In this paper we propose a graphic-processing-unit (GPU)-based regular LDPC decoders with the log sum-product iterative decoding algorithm (log-SPA). The GPU code was written to run NVIDIA GPUs using the compute unified device architecture (CUDA) language with a novel implementation of asynchronous data transfer for LDPC decoding. Experimental results showed that the proposed GPU-based high-throughput regular LDPC decoder achieved a significant 271x speedup compared to its CPU-based single-threaded counterpart written in the C language. [ABSTRACT FROM PUBLISHER]

Software based decoding of low-density parity-check (LDPC) codes frequently takes very long time, thus the general purpose graphics processing units (GPGPUs) that support massively parallel processing can be very useful for speeding up the simulation. In LDPC decoding, the parity-check matrix H needs to be accessed at every node updating process, and the size of the matrix is often larger than that of GPU on-chip memory especially when the code length is long or the weight is high. In this work, the parity-check matrix of cyclic or quasi-cyclic (QC) LDPC codes is greatly compressed by exploiting the periodic property of the matrix. Also, vacant elements are eliminated from the sparse message arrays to utilize the coalesced access of global memory supported by GPGPUs. Regular projective geometry (PG) and irregular QC LDPC codes are used for sum-product algorithm based decoding with the GTX-285 NVIDIA graphics processing unit (GPU), and considerable speed-up results are obtained. [ABSTRACT FROM AUTHOR]

This paper proposes the implementation of a zero-order Takagi–Sugeno–Kang (TSK)-type fuzzy neural network (FNN) on graphic processing units (GPUs) to reduce training time. The software platform that this study uses is the compute unified device architecture (CUDA). The implemented FNN uses structure and parameter learning in a self-constructing neural fuzzy inference network because of its admirable learning performance. FNN training is conventionally implemented on a single-threaded CPU, where each input variable and fuzzy rule is serially processed. This type of training is time consuming, especially for a high-dimensional FNN that consists of a large number of rules. The GPU is capable of running a large number of threads in parallel. In a GPU-implemented FNN (GPU-FNN), blocks of threads are partitioned according to parallel and independent properties of fuzzy rules. Large sets of input data are mapped to parallel threads in each block. For memory management, this research suitably divides the datasets in the GPU-FNN into smaller chunks according to fuzzy rule structures to share on-chip memory among multiple thread processors. This study applies the GPU-FNN to different problems to verify its efficiency. The results show that to train an FNN with GPU implementation achieves a speedup of more than 30 times that of CPU implementation for problems with high-dimensional attributes. [ABSTRACT FROM AUTHOR]

The Moving Particle Semi-implicit (MPS) method performs well in simulating violent free surface flow and hence becomes popular in the area of fluid flow simulation. However, the implementations of searching neighbouring particles and solving the large sparse matrix equations (Poisson-type equation) are very time-consuming. In order to utilize the tremendous power of parallel computation of Graphics Processing Units (GPU), this study has developed a GPU-based MPS model employing the Compute Unified Device Architecture (CUDA) on NVIDIA GTX 280. The efficient neighbourhood particle searching is done through an indirect method and the Poisson-type pressure equation is solved by the Bi-Conjugate Gradient (BiCG) method. Four different optimization levels for the present general parallel GPU-based MPS model are demonstrated. In addition, the elaborate optimization of GPU code is also discussed. A benchmark problem of dam-breaking flow is simulated using both codes of the present GPU-based MPS and the original CPU-based MPS. The comparisons between them show that the GPU-based MPS model outperforms 26 times the traditional CPU model. [ABSTRACT FROM AUTHOR]

The Dey-Mittra conformal finite difference time domain (CFDTD) algorithm for perfect electrical conductors is modified for the analysis of finite conductivity conductors at millimeter wave frequencies. Formulas are derived for CFDTD coefficients using voltage state variables and a constant surface impedance boundary condition (SIBC). The approach permits a fast implementation suitable for CUDA type GPU hardware. Accuracy and stability are investigated with respect to the stability constraints on intersection areas introduced by Dey-Mittra and Benkler as well as the distance stability constraints of Zagorodnov that permits 100% Courant temporal sampling. A relaxation of the Zagorodnov distance constraints permits increased accuracy with respect to all alternative area constraints. Analytic solutions are used to judge the performance of the proposed CFDTD modifications for millimeter wave band applications. [ABSTRACT FROM AUTHOR]

Visualized freehand 3-D ultrasound reconstruction offers to image incremental reconstruction during acquisition and guide users to scan interactively for high-quality volumes. We originally used the graphics processing unit (GPU) to develop a visualized reconstruction algorithm that achieves real-time level. Each newly acquired image was transferred to the memory of the GPU and inserted into the reconstruction volume on the GPU. The partially reconstructed volume was then rendered using GPU-based incremental ray casting. After visualized reconstruction, hole-filling was performed on the GPU to fill remaining empty voxels in the reconstruction volume. We examine the real-time nature of the algorithm using in vitro and in vivo datasets. The algorithm can image incremental reconstruction at speed of 26–58 frames/s and complete 3-D imaging in the acquisition time for the conventional freehand 3-D ultrasound. [ABSTRACT FROM PUBLISHER]

The shooting and bouncing ray (SBR) method is highly effective in the radar cross section (RCS) prediction. For electrically large and complex targets, computing scattered fields is still time-consuming in many applications like range profile and ISAR simulation. In this paper, we propose a GPU-based SBR that is fully implemented on the graphics processing unit (GPU). Based on the stackless kd-tree traversal algorithm, the ray tube tracing can rapidly evaluate the exit position in a single pass on the GPU. We also present a technique for fast electromagnetic computing that allows the geometric optics (GO) and Physical optics (PO) integral to be carried out on the GPU efficiently during the ray tube tracing. Numerical experiments demonstrate that the GPU-based SBR can significantly improve the computational efficiency of the RCS prediction, about 30 times faster, while providing the same accuracy as the CPU-based SBR. [ABSTRACT FROM AUTHOR]

Edge computing represents the future of computing paradigms that perform tasks near the user plane. The integration of blockchain with edge computing provides the added advantage of secured and trusted communication. In this paper, we propose a blockchain-based service migration by developing edge clusters using NVIDIA Jetson boards in an embedded edge environment, using containers and Kubernetes as a container orchestration capable of handling real-time computation-intensive deep learning tasks. Resource constraints in the edge and client movement are the proposed scenarios for service migration. Container migration due to mobile clients is integrated with blockchain to find a suitable destination, meta-based node evaluation, and secured data transfer in the connected car environment. Each service request migration takes, on average, 361 ms. The employed container migration method takes 75.11 s and 70.46 s to migrate application containers that use NVIDIA CUDA Toolkit. Finally, we evaluate the efficiency of blockchain to find the destination node through performance parameters such as latency, throughput, storage, and bandwidth. [ABSTRACT FROM AUTHOR]

The Utah Energy Balance (UEB) model supports gridded simulation of snow processes over a watershed. To enhance computational efficiency, we developed two parallel versions of the model, one using the Message Passing Interface (MPI) and the other using NVIDIA's CUDA code on Graphics Processing Unit (GPU). Evaluation of the speed-up and efficiency of the MPI version shows that the effect of input/output (IO) operations on the parallel model performance increases as the number of processor cores increases. As a result, although the computation kernel scales well with the number of cores, the efficiency of the parallel code as a whole degrades. The performance improves when the number of IO operations is reduced by reading/writing larger data arrays. The CUDA GPU implementation was done without major refactoring of the original UEB code, and tests demonstrated that satisfactory performance could be obtained without a major re-work of the existing UEB code. • The Utah Energy Balance snowmelt model computational kernel is highly parallelizable. • Input/Output was the major efficiency bottleneck in parallelization. • GPU implementation was achieved without major refactoring of existing UEB code. [ABSTRACT FROM AUTHOR]